Robust visual performance of Bsn-/- mice

In the present study, we used Bassoon mutant mice, which display reduced signal transfer from photoreceptor to bipolar cells (Dick et al., 2003) to study the central processing and visual capabilities of animals with disturbed retinal signals. In Bsn-/- mice, neuronal transmission from photoreceptor to bipolar cells is impaired. Electroretinographic recordings revealed that the b-wave, which represents activity in 2^nd order retinal neurons, was diminished and had a slower time course in Bsn-/- mice (Dick et al., 2003). While these results showed that there is residual synaptic transmission from photoreceptors to bipolar cells in the retina of Bsn-/- mice, they did not show to which extent the remaining neurotransmission is used for the activation of the central vision system and for behavioral tasks.

To address these questions mice were behaviorally tested in two different paradigms: the virtual-reality optomotor system (Prusky et al., 2004) and the visual water task (Prusky et al., 2000). In addition, visual cortical maps were recorded by optical imaging of intrinsic signals (Kalatsky and Stryker, 2003).

Visual acuity measured with the virtual-reality optomotor system was significantly reduced in mutants compared to littermate controls but also reached adult values already at four weeks of age. In the visual water task, both Bsn+/+ and Bsn-/- mice had higher visual acuity than in the virtual-reality optomotor system. This difference was previously observed for C57Bl/6J mice (Prusky et al., 2000) and is most likely due to different neuronal subsystems subserving the behavioral responses in optomotor and reinforcement-based tasks:

optokinetic movements are driven by subcortical low-frequency visual pathways while grating acuity in the visual water task is driven by cortical circuits (Prusky et al., 2000; Prusky et al., 2004; Douglas et al., 2005).

Interestingly, the difference in visual acuity between mutants and controls was identical in both tasks: Bsn-/- mice had visual thresholds about approximately 0.2 cyc/deg lower than Bsn+/+ mice. Our results indicate that mice can achieve a visual acuity of about 0.4 cyc/deg

127 even in the absence of proper ribbon synapses and that Bassoon-dependent fast exocytosis is necessary “only” for the additional 0.2 cyc/deg observed in wild-type mice.

The contrast sensitivity of the mutants was notably reduced in the optomotor test, while mutants and wild-type mice performed similarly in the optical imaging experiments. This interesting result may be explained by a compensation of lost contrast enhancing mechanisms in the retina by contrast enhancing mechanisms in more central parts of the visual system which are not involved in mediating optomotor reflexes. Conversely, the reason for the reduction of contrast sensitivity in the optomotor task in Bsn-/- mice may be that the retinal signals for the optomotor reflexes are processed by subcortical pathways, so that a loss of contrast enhancing mechanisms in the retina cannot be compensated for by central parts of the visual system. A possible explanation for reduced contrast sensitivity mechanisms in the retina of Bsn-/- animals may be based on the morphology of the ribbon synapse. In this particular synapse, the postsynaptic dendritic contacts of horizontal cells seem to embrace the anchor site of the ribbons, whereas the contact sites of the bipolar cells are somewhat more distant. This morphology makes it reasonable to assume that a loss of ribbons may affect the synaptic transfer between photoreceptors and horizontal cells more severely than the transfer between photoreceptors and bipolar cells and therefore reduce the efficiency of the inhibitory periphery of the receptive fields. Such a reduction of the inhibitory interactions should result in a reduction of both contrast sensitivity and spatial resolution, as we have indeed observed in our experiments.

Both of our behavioral tests clearly showed that ribbon-dependent fast exocytosis from photoreceptor cells is essential for normal visual capabilities. These data also confirm previous observations that synaptic ribbons are essential for the precise information processing at photoreceptor synapses as well as at inner ear hair cell synapses and that active zones without ribbons still retain sufficient functional performance to mediate sensory signal transduction reaching central systems (Dick et al., 2003; Khimich et al., 2005).

The possibility that the reduced visual capabilities of Bassoon mutant mice are at least in part due to the function of the Bassoon protein in synapses of the cerebral cortex cannot be excluded (Altrock et al., 2003; Angenstein et al., 2007; Angenstein et al., 2008).

Furthermore, since the range of the inhibitory system is tuned to the optimal spatial frequency (Sagi and Hochstein, 1985; Shapley and Lennie, 1985), it can be expected that the

128 reduction of contrast sensitivity is largest at this particular spatial frequency. This was also observed in our experiments in the virtual-reality optomotor system: the reduction of contrast sensitivity was highest at the optimal spatial frequency of 0.064 cyc/deg (six times) and much less (two to five times) at other spatial frequencies. It is therefore possible that contrast enhancing mechanisms in the retina (Kamermans and Spekreijse, 1999) are reduced or switched off while the projection via the bipolar cells is still more or less intact. A reduced input to retinal horizontal cells could also explain the alterations in the b-wave of the electroretinogram in Bassoon mutants (Dick et al., 2003). A reduced and prolonged b-wave could result when the inhibitory system of the horizontal cells is switched off. Thus, the absence of activity in horizontal cells could account for the reduced amplitude of the b-wave and the absence of inhibitory interactions could explain its prolonged duration.

Interestingly, our tests in the optomotor system with different stimulus velocities revealed a significant difference in the temporal sensitivity between wild-type and mutant animals:

Bsn-/- mice tracked grating stimuli up to 38 °/sec while littermates continued to track the gratings to over 50 °/sec, the technical speed limit of our system. These results are further evidence for decelerated neuronal processing in the visual pathway of Bsn-/- mice, presumably at early stages of synaptic processing in the retina, as it is indicated by the prolonged duration of the ERG b-wave (Dick et al., 2003). While the residual synaptic transmission from photoreceptors to bipolar cells is sufficient to mediate basic visual capabilities for Bsn-/- mice, a deficit becomes obvious when the system is challenged with higher stimulus velocities.

Surprisingly, we found that optical imaging of visual cortical activity (Kalatsky and Stryker, 2003) revealed essentially no differences between Bsn-/- mice and wild-type littermates.

Cortical maps were also adult-like at four weeks of age. These results show that i) robust visual performance can be achieved in absence of synaptic ribbons, ii) development and maintenance of visual cortical maps are independent of synaptic ribbons and iii) visual development of Bsn-/- mice is completed at four weeks of age, which indicates that later developing ectopic synapses (Dick et al., 2003; Specht et al., 2007) do not seem to contribute to visual capabilities nor improve central visual processing. Hence, the central visual system can make use of slow and weak retinal signals from a diseased retina to subserve surprisingly robust vision.

129 Furthermore, optical imaging of intrinsic signals in the visual cortex did not reflect the impairment of the afferent visual signal, but must have been amplified either by some intrinsic cortical processing or at some stage of the afferent visual pathway beyond the defective retinal transmission. Thus, retinal ganglion cells or more central stages of visual information processing are obviously able to compensate a loss of lateral inhibition in the retina.

In the absence of functional Bassoon, ectopic photoreceptor synapses are formed in the outer nuclear layer of the retina starting at about four weeks postnatally (Dick et al., 2003).

By examining mutant retinae over a period of two years long-lasting plastic changes were observed: ectopic photoreceptor synapses were formed de novo by sprouting of both bipolar and horizontal cells resulting in a dense synaptic plexus in the outer nuclear layer, which is absent in wild-type retina. To check whether ectopic synapses are responsible for our observations in adult Bsn-/- mice, we followed vision longitudinally in individual animals from eye opening and also imaged visual cortical maps. These experiments revealed that Bsn-/- mice reached adult values of both visual acuity and contrast sensitivity at four weeks of age and thus before ectopic synapse formation modified retinal structure. In addition, visual cortical maps were adult-like at the same age. It is therefore highly unlikely that the formation and remodeling of ectopic synapses, which lasts for at least one year (Specht et al., 2007) is responsible for the visual abilities of Bsn-/- mice or improves vision or central visual processing. It rather indicates that the remaining neuronal transmission of the photoreceptor synapse in mutant mice, that is basic exocytosis at ribbon-deficient synapses and additional fusion outside the active zone (Dick et al., 2003; Khimich et al., 2005), is sufficient to sustain decent visual capabilities and activate central visual processing structures.

Taken together, our analyses showed that while Bassoon-dependent fast exocytosis is essential for normal visual capabilities behaviorally relevant visual discrimination was still possible in Bsn-/- mice.

Our results are consistent with a previous report showing that the development of precise retinotopic maps in the visual cortex of mice does not depend on experience-dependent sensory input but is due to waves of spontaneous retinal activity during the first postnatal week. By using a mutant mouse line that has uncorrelated retinal activity during

130 development due to the deletion of the β2-subunit of the nicotinic acetylcholine receptor it was demonstrated that the geniculocortical map reaches adult precision by postnatal day eight and is not further refined by later retinal waves or visual experience (Cang et al., 2005b).

Furthermore, our optical imaging results show that the slower and reduced input signals from the Bassoon mutant retina into visual cortex are not only sufficient to sustain retinotopic maps but also to activate the cortex as strongly as in normal mice. We take this observation as a strong indication of homeostatic mechanisms (Turrigiano and Nelson, 2004). While it is generally believed that Hebbian type of network modifications is crucial for shaping cortical circuits after sensory experience, additional homeostatic plasticity might ensure that network compensation can be achieved in spite of a wide range of sensory perturbations (Mrsic-Flogel et al., 2007; Maffei and Turrigiano, 2008). We suggest that in Bsn-/- mice, homeostatic synaptic scaling might had adjusted excitatory synaptic strengths of neurons in the afferent visual pathway or in the visual cortex to compensate for reduced synaptic transmission in the retina.

Overall our results show that the central visual system has an extraordinarily high potential to process altered inputs from the retina and hence to be instrumental in sustaining behaviorally relevant visual capabilities (Goetze et al., 2010b).